LAPPEENRANTA UNIVERSITY OF TECHNOLOGY School of Business and Management
Master in Strategic Finance and Business Analytics
Benjamin Rencz
ENHANCING MANAGERIAL CONTROL IN PRODUCTION ECONOMICS THROUGH ANALYTICS:
CASE OF GLOBAL SPARE PARTS SUPPLY CHAIN MANAGEMENT
Examiners: Professor Mikael Collan, LUT
Robert Jenks, Director of Global Spare Parts Operations
ABSTRACT
Author: Rencz, Benjamin Jacob
Title: Enhancing managerial control in production economics through analytics: case of global spare parts supply chain management
School: School of Business and Management Master’s Programme: Strategic Finance and Business Analytics
Year: 2017
Master’s Thesis: Lappeenranta University of Technology,
80 pages, 13 figures, 15 tables and 10 appendices Examiners: Professor Mikael Collan LUT
Robert Jenks, Director of Global Spare Parts Operations Keywords: SCOR Method, Spare Parts, SCM, Decision Making, Analytics, KPI
This thesis focuses on the interface of production economics and business analytics to support managerial control. Specifically, within the field of production economics the thesis deals with the study of spare parts supply chain management (SCM) for a global spare parts company and the utility of analytics, analyzing past performance to enhance managerial control through performance measurement. A review of the literature reveals that limited research has been conducted on the impact of integrating analytics and Big Data in global spare parts SCM. Based on the available research, the thesis develops a methodology for enhancing the utility of performance measurement for managerial control. A single case approach is utilized in the research. Big Data, compiled from several of the case company’s databases, is used within a Geographic Information System and Supply Chain Operational Reference (SCOR) framework to demonstrate that benchmarking would provide valuable insight into the existing business processes of spare parts SCM. It also demonstrates that the use of SCOR metrics provides a more comprehensive understanding of the case company’s Key Performance Indicators (KPI). Finally, the research finds that the use of benchmarking provides greater clarity concerning trends in current SCM processes for managerial control.
ACKNOWLEDGEMENTS
As I reflect back upon the process of writing this thesis, I would be remiss if I didn’t acknowledge the individuals who played important roles throughout this work. First, I wish to thank Robert and Anselmi, who trusted in my ability to develop the ideas we discussed and turn them into what I hope is something useful for KONE. The guidance I received from the entire GSS unit throughout was invaluable. My many colleagues and coworkers, Hilkka, Jani, Nnamdi and Tamara, who always made time to discuss the thesis. I also want to recognize my supervising professor, Professor Mikael Collan, for his guidance and patience in reviewing and commenting on the many versions of this thesis. Finally, I know that I would not be where I am today without the ongoing support of my parents and, in particular, without Tuuli, who is the wind in my life encouraging and supporting me forward.
Hyvinkää, March 11, 2017 Ben Rencz
TABLE OF CONTENTS
1 INTRODUCTION ... 9
1.1 Background ... 10
1.2 Motivation for the research ... 12
1.3 Research Aim ... 13
1.4 Literature Review ... 13
1.5 Research Questions ... 18
1.6 Research Methodology ... 20
1.7 Research Structure ... 21
2 METHODOLOGY ... 22
2.1 Single Case Study ... 22
2.2 Supply Chain Performance Measurement Framework ... 22
2.3 Three Research Phases ... 24
2.3.1 Phase 1: “As Is” Description ... 24
2.3.2 Phase 2: Metrics’ Relationships... 25
2.3.3 Phase 3: Performance Enhancement ... 26
3 CASE: SPARE PARTS SUPPLY CHAIN MANAGEMENT OF A MULTINATIONAL COMPANY ... 28
3.1 Case Company: KONE Corporation ... 28
3.1.1 Global Spares Supply unit of the Case Company ... 28
3.1.2 Defining the Business Structure in Global Spares Supply ... 30
3.1.3 Case Company Needs ... 31
3.2 Data Analysis ... 31
3.2.1 Data Compilation ... 31
3.2.2 Statistical Characterization ... 33
3.2.3 KPI Calculation, Global Spare Supply Unit ... 34
3.2.4 Geographic Information System Analysis ... 35
3.2.5 SCOR Metrics Calculation ... 36
3.3 Validity and Reliability of the Case ... 39
3.3.1 Reliability ... 39
3.3.2 Validity ... 40
4 RESULTS ... 41
4.1 CPI Characterization ... 43
4.2 KPI and CPI Statistical Characterization ... 48
4.3 Relationship between KPI and CPI ... 49
4.4 SCOR Performance Metrics ... 57
4.5 Validation of Metrics ... 60
5 DISCUSSION ... 63
5.1 Research Question One ... 64
5.2 Research Question Two ... 64
5.3 Research Question Three ... 65
6 CONCLUSION ... 67
6.1 Summary of the Contributions ... 67
6.2 Recommendations to Management ... 69
6.3 Limitations ... 71
6.4 Summary ... 71
REFERENCES ... 73
APPENDICES ... 81
Appendix 1. Results from query used in literature review ... 82
Appendix 2. Description of Supply Chain Frameworks for creating performance measurement metrics following criteria presented by Piotrowicz (2015) ... 83
Appendix 3. Relational data model (Peter 2011) ... 87
Appendix 4. Distribution curves for performance metrics KPI and CPI ... 88
Appendix 5. Distribution curves for log transformed KPI values ... 89
Appendix 6. Case KPI Basic Statistics ... 90
Appendix 7. Rational for removal of outliers with SCOR metric calculation. ... 91
Appendix 8. Tests for normality of SCOR metrics ... 92
Appendix 9. Bartlett test for SCOR Metrics ... 93
Appendix 10. Bartlett Test for SCOR metrics with Cox Transformation ... 94
LIST OF FIGURES
Figure 1. Research focus: Interface of production economics with analytics and
managerial control ... 9
Figure 2. Methodological approach to address research questions ... 24
Figure 3. Overview of the GSS offices and performance levels for 2015 ... 29
Figure 4. Supply chain flow for the four units within Global Spares Supply ... 30
Figure 5. Three critical process steps in calculating SCOR metrics and preparation for Data Science ... 36
Figure 6. Characteristics of SCOR metrics as calculated in study... 37
Figure 7. Distribution of X countries receiving spare parts from Country D, Country C or Country A ... 44
Figure 8. Shipping distribution for the three-major spare part producing countries: Country D (top), Country C (mid) and Country A (bottom) ... 46
Figure 9. Ratio of CPI for the X countries receiving spare parts from Country D and Country C ... 47
Figure 10. CPI Ratio of Shipping Condition 1 vs. Shipping Condition 2 for Country D Supplier Parts ... 48
Figure 11. 3D Scatter plot of Principal Component Analysis with case SCOR metrics ... 53
Figure 12. 3D Scatter plot of Principle Component Analysis with case KPIs ... 55
Figure 13. Overall strength of KPI (blue) and SCOR (red) measurements as indicators for company performance. Arrow direction illustrates greater utility of a metric ... 62
LIST OF TABLES
Table 1. Structured literature review results ... 15
Table 2. Critical criteria for selecting a SCM performance metric in spare parts ... 17
Table 3. Twelve key performance indicators used in the Global Spare Supply unit ... 35
Table 4. SCOR Metric Composition ... 38
Table 5. Description and function of the four units within the Global Spare Supply unit ... 42
Table 6. Shipping networks for spare parts ... 43
Table 7. CPI values for major producing countries ... 45
Table 8. Correlation coefficients for KPIs and CPI (ESP) ... 51
Table 9. PCA results and loadings for SCOR metrics ... 52
Table 10. PCA values for case company KPIs ... 54
Table 11. Multiple linear regression results for team KPIs vs. CPI ... 56
Table 12. SCOR Metric Basic Statistics ... 58
Table 13. SCOR metric Pearson correlation coefficients where r2 > .5 is strong (green) and r2 < .3 is weak (grey) ... 59
Table 14. Linear Regression for SCOR Metric vs. CPI ... 60
Table 15. Contributions to the field of PM Spare Parts SCM ... 69
GLOSSARY OF TERMS
CoV Covariance
CPI Core Performance Indicator CS Customer Service Team
GIS Geographic Information Systems GSS Global Spare Supply
IN Invoicing Team ISCO Inventory Team
KPI Key Performance Indicator MC Managerial Control
MM Material Management Team PCA Principal Component Analysis PE Production Economics
PM Performance Measurement SCM Supply Chain Management
SCOR Supply Chain Operations Reference SCPM Supply Chain Performance Measurement VIF Variance Inflation Factors
1 INTRODUCTION
This thesis focuses on the interface of production economics and analytics to support managerial control. Figure 1 outlines the focus of the thesis. Specifically, within the field of Production Economics the thesis deals with the study of spare parts Supply Chain Management within a large multi-national company and the utility of Analytics, to analyze past performance, to enhance Managerial Control through Performance Measurement.
Figure 1. Research focus: Interface of production economics with analytics and managerial control
Production Economics
- Supply Chain Management
Managerial Control
- Performance Measurement
Analytics
- Business Analytics
Research Focus
- Spare Parts SCM - Decision Making - KPI
1.1 Background
The development of useful Performance Measurements is fundamental to the success of companies as they provide ongoing feedback for Managerial Control (Gunasekaran 2001; Vera-Baquero 2013). However, the development of valid indicators is complicated and requires an integrated knowledge of several disciplines as shown in Figure 1 (Mishra 2014). These disciplines need to be understood and placed in the context of the specific needs of a company.
Production Economics focuses on the combination of engineering and management.
The primary topics in the domain are: production, manufacturing and process industries. The ultimate goal of the discipline is to enhance knowledge of industry practice to support managerial control. A significant component of production economics is supply chain management. (Kotler et al. 2006)
Supply chain management (SCM) is monitoring and controlling of goods and services between the point of origin to the point of use, involving every company in at least some aspect (Christopher 2016). SCM is an integral component of operations in large corporations as it provides the link between and within corporations. These external and internal linkages supported with robust data, clarity of findings and insights into their performance provide organizational strength. Spare Parts SCM includes both the products and the services that are required to maintain the original operating condition of the equipment for the end user. (Botter & Fortuin 2000)
Managerial Control refers to formal and informal mechanisms, processes, systems and networks used by management that involve setting, monitoring and controlling performance standards as well as methods. The output of managerial control is exhibited through the value produced from the corrective actions of the managers.
These actions are enhanced through the utilization of appropriate performance measurements. (Grubbström 2017)
Performance Measurement (PM), based upon a series of indicators or metrics, is the process of gathering, examining and reporting information against companies’
strategic objectives to understand the performance of internal or external stakeholders, processes or systems (Mishra 2014). These PMs are directed and supported through analytics (Grubbström 2017).
Analytics is the methodical computational analysis of collected historical events that are represented as data. The data can incorporate multidimensional fields that use mathematics, statistics, behavior modelling to provide meaningful results and visualization. (SAS 2017) Business analytics is the complex set of statistics used to analyze business data for the purpose of making a business decision. Business Analytics methods are a valuable component in a PM process, providing the analysis of the performance indicators. (Davenport 2009)
Combining business analytics and PMs allows for a variety of methods to measure success in SCM. Supply Chain Analytics (SCA) is simply using business analytics in the context of SCM as identified by Wang and Gunasekaran (2016). SCA is used in the development of Key Performance Indicators (KPIs) to enhance their utility as comprehensively measuring numerous processes (Vera-Baquero 2016). A KPI is a type of PM that organizations use to evaluate the overall success of a specific strategic process. This singular measure becomes a proxy for a varied and complex set of processes involved in a supply chain. (Wang & Gunasekaran 2016)
Supply Chain Performance Management (SCPM) frameworks are being used to enhance the utility of KPIs for business operations and to enhance managerial control.
Although the use of KPI’s is accepted there are issues concerning the validity of these metrics and therefore their utility in support of managerial control. Recent technological advances provide an opportunity to validate the metrics and to enhance their application. This directly speaks to mature methods in business analytics and the use of “Big Data”. Big Data can be roughly stated as data that are too large in terms of volume and complexity for traditional techniques and analysis methods.
These mature capabilities coupled with Big Data’s ability to numerically capture a complex spare parts SCM process may be useful in understanding and monitoring business processes. (Gandomi 2015)
1.2 Motivation for the research
Spare parts SCM processes are often massive in size and differ greatly depending on the context of the business. The spare parts industry continually experiences pressures with new trends routinely coming into scope. Spare parts also has a related requirement of managing the challenges of a varied and ever increasing spare parts inventory with geographically sensitive distribution networks.
Consequently, the challenge for spare parts SCM is to address the inherent complexity of their processes in addressing these variations, while meeting the internal needs of the organization for managerial control. (Gawankar 2016; Tyagi 2015) In other words, there is a need to identify the difference between the planning, execution and delivery of a company’s products or services, highlighting variances and potential areas for improvement. This requires a robust and sensitive PM process for managerial control. The majority of global companies spend considerable resources on the creation of PM systems and processes. Further, PMs have become fundamental business practices for all players involved in spare parts SCM. (Chan 2009)
In PM systems, KPIs are used as a proxy for a varied and complex set of processes involved in SCM as opposed to using a diverse set of multiple indicators, to simplify the reporting process for managerial control. However, the analytical approaches to provide this broad overview have become increasingly complicated, often with the result that it is unclear what the KPI means. In other words, a manager could have a KPI in spare parts SCM that appears to be failing, status red, and yet the underlying indicators all appear to be strong, status green. Obviously, the design of the KPI or the aggregation of information used in aggregating the “sub-indicators” to a single KPI fails in cases like the above, leading to an undesirable outcome. This leads to
concerns with the validity of KPIs. As a result, PMs have not always been implemented as effectively in spare parts SCM. (Wang & Gunasekaran 2016)
To address this issue of validity and utility of SCM KPIs, Chan (2009) suggests that a simpler approach should be explored. He stresses the importance of understanding the underlying business processes to help achieve successful spare parts SCM. Supply Chain Performance Management (SCPM) Frameworks have been developed to address this issue and to facilitate the linkage between business processes and PMs into a unified structure. The use of SCPM Frameworks enhances the uptake of KPIs into business operations. Recent technological advances provide an opportunity to enhance this further. This directly speaks to mature methods in business analytics and the use of Big Data. These mature capabilities coupled with Big Data’s ability to numerically capture a complex spare parts SCM process may be useful in understanding and monitoring business processes. (Balfaqih et al. 2016; Gandomi 2015; Tyagi 2015)
1.3 Research Aim
The main aim of this research is to enhance the utility of performance indicators in spare parts SCM using quantitative analysis. The academic merits lie in an increased understanding of the variability in spare parts SCM as related to the proper use of PM.
The corporate implication is that managers would have better information for managerial control on SCM performance issues. This will enhance managers’
understanding of KPI behavior and facilitate the utilization of performance measurement. (Tyagi 2015)
1.4 Literature Review
A literature review is conducted to identify critical components and research gaps that would support development of a research topic in development of PM within a spare part SCM. The literature review for this thesis starts with three main concepts:
production economics, analytics and managerial control. To explore the concepts, a basic query, as shown in Appendix 1, was executed over several academic databases. The initial query of the main concepts of Managerial Control and Production Economics yielded no relevant results. Next the sub-topics from the main concepts were chosen for the query key words: Supply Chain Management replaced Production Economics, Performance Measurement replaced Managerial Control and Analytics remained. These query inputs were combined with the main research areas of the thesis, which are spare parts and KPI, over a multiple database search. As a result of a multiple database search, as outlined in Appendix 1, “Indersciece” was identified as the preferred database. Table 1 shows the results from the
“Inderscience” database, including: author, query words, date of article retrieval, query results and criteria for selection. The primary topic of the literature review, SCM, wasn’t used in the query search directly due to it adding too broad a scope to the query. Rather, the query results’ abstracts were examined if they were framed around SCM. Table 1 shows the five articles that result from the structured literature review.
Table 1. Structured literature review results
Author
Selected Kasi Lin &
Ghodrati
Beekman &
de Leeuw Velimirović Öberg
Article Title
Systemic assessment of SCOR for modeling supply chains
Maintenance spares inventory management performance measurement using a HOMM
Supply chain oriented performance measurement for
automotive spare parts
Role and importance of key
performance indicators measurement
Changing from watermelon measures to real decision support:
including information about variation in
performance measurements
Query Search
spare parts AND
Performance Measurement AND KPI
spare parts AND KPI AND
Performance Measurement
Performance Measurement AND Spare Parts
KPI AND spare parts AND
Performance Measurement
KPI AND Analytics AND Decision Making Search
Field All fields All fields All fields All fields All fields Time
Span All years All years All years All years All years
Results 3 5 19 8 2
Criteria for selection
Spare parts and
International Context
Selected Highest Relevance
Focus on Spare parts and PM
Discusses decision making &
indirectly spare parts
Interpretation of findings of KPI selection in SCM, use of analytics
Date
Retrieved 12.02.2017 12.02.2017 13.02.2017 13.02.2017 15.02.2017
All of these papers identify the need for more study in the area of spare parts SCM.
The majority of the articles cite Gunasekaran’s (2001, 2004) research, where it is noted that an important area for further study would be the selection of an appropriate SCPM framework. However, Gunasekaran (2004) also notes that this is complicated
by the vast number of options and the challenge of selecting the suitable model to address the needs of a particular company.
Lin and Ghodrati (2012), in a study of spare parts management identify significant gaps in the utility of PM in spare parts. Specifically, they consider that KPIs are inadequately built for the unique demands of spare parts in a global context, particularly with reference to both its technical complexity and social context. The researchers attempt to address the gap by developing a framework called HOMM (House of Maintenance Management). They report some success. However, they note that the framework is dependent upon a three-year implementation which does not meet most company requirements of timely indicators (Vera-Baquero 2013 2016).
Kasi (2007) recommends the Supply Chain Operations Method (SCOR) framework to aid in managerial control for global SCM. The paper identifies several aspects that they consider lacking in Lin and Ghodrati’s (2012) study but which are contained in the SCOR framework. They note the strength of the SCOR framework lies in its technical dimensions such as its modeling process capabilities and clear operating instructions for implementation. However, the paper notes the lack of social dimensions in the SCOR mode and recommends further testing of the model in a wider variety of industrial environments to confirm its overall strength, suggesting a spare parts environment given its inherent complexities. (Kasi 2007; Lin & Ghodrati 2012)
De Leeuw and Beekman (2008) study PM in automotive spare parts SCM environments comparing two frameworks, SCOR and LogistiQual. Using qualitative study methods, they investigate the validity of the KPIs, specifically whether the KPIs effectively measure the underlying business processes. They find that the LogistiQual versus the SCOR is more facile in the automotive spare parts environment while still noting the value of the SCOR framework in spare parts. There appears to be some
consensus that the SCOR framework is an appealing model because it is both a structured and manageable approach.
A challenge identified by all these studies is the selection of the appropriate KPI for use within a given SCPM framework. Each SCPM framework has numerous KPIs.
The challenge for a company is to know which KPI is best suited for their needs.
Some KPIs are more holistic, while others are more detailed. Every approach has its tradeoffs. For example, a holistic indicator will give a broad overview but it may not provide the granularity of reporting that is required to make an operational decision.
While many authors present useful insights, de Leeuw and Beekman (2008) provide the most comprehensive list of criteria for the selection of a KPI for a spare parts company, as noted in Table 2.
Table 2. Critical criteria for selecting a SCM performance metric in spare parts (de Leeuw & Beekman, 2008)
# Criteria for spare parts SCM KPI
1. Stock turnover rate, inventory process time, total stock turnover, cycle time 2. Mean cost per line, mean cost per movement
3. Complete inventory value, inventory balances
4. Average service level, stock accessibility, fill rate time
5. Delivery accuracy, documenting accuracy, forecasts accuracy 6. Punctual delivery performance at requested time and place
7. Customer satisfaction level, complaint percentage, degree of failures
The next aspect identified by the studies is the database used for the measurement of the KPI. Velimirović (2011) examines data selection for calculating KPIs in spare parts SCM. It simplified the process by sub-dividing the data into two areas: Non- Financial and Financial. This is a simpler approach than previous studies where an overall framework is investigated (Lin & Ghodrati 2012; Kasi 2007; de Leeuw &
Beekman 2008). Velimirovic (2011) argues that subdividing the data to better reflect the relevant underlying business processes that a KPI is measuring, would increase the validity of the KPI.
A paper by Oberg et al. (2016) focuses on the validity of KPIs in PM of spare parts SCM. The study uses graphic design to visually demonstrate how data changes over a given period of time, which enhances managers’ ability to understand where corrective action could be taken. Further, the study finds that if the data are divided into sub-categories that reflect patterns in business processes, managers would be enabled to act on the data more confidently. This finding is in line with de Leeuw and Beekman’s (2008) paper which recommends testing data over periods of time and space for international companies. It is stated that this increased the validity of the KPI and thereby, its utility for managerial control. Another key finding of the study is the impact of geographic differences on PM processes. KPIs appear to be responsive to unidentified geographic variation in data and that certain valuable trends and patterns could be concealed. The aspect of geographic variation is relatively important, yet with limited research on its impact in the area of PM in spare parts SCM.
In summary, the literature clearly identifies the need for more study in the area of spare parts SCM PM. Within that scope, the review identifies issues of KPI validity and the resulting lack of clarity concerning PM results for managerial control. The literature does note some promising areas for study, specifically: the use of a SCPM framework; KPI selection criteria; data development for KPIs; subdivision of data geographically and longitudinally. The research questions and methodology are developed based upon these findings.
1.5 Research Questions
Based upon the research aim and the results of the literature review, this thesis addresses three research questions.
1: Can the use of an industry standard performance measurement framework enhance the control of spare parts SCM for a case company?
The SCMP framework produces a metric that monitors performance aspects within a company. An effective metric will flag an issue and support managerial control to address the issue. An initial step in developing the metric is the selection of an appropriate framework that aligns with a company needs. The selection requires an assessment of available frameworks as well as company needs. Subsequently the actual calculation of the metric and its implementation need to adhere to proper protocol to ensure the utility of the metric in managerial control.
2: Can the use of business analytics, including the use of Big Data and new technologies, such as geographic information systems, support the process of developing useful performance metrics for a case company?
The process to create KPIs is dependent upon data and, if the analysis is meant to provide robust and interpretable results, data quality must be assured. A business analytics approach will be used to characterize the data, investigate the quality of data and investigate the relationships between the metrics. Big Data will be included in this evaluation. The ability of new technologies that map geographic patterns will also be used to address the spatial variation in the data and how this impacts development of useful PMs. These new tools could support the new trends in spare parts SCM by providing real time indicators and the requirement for a holistic overview of the business environment for the case company in this study.
3: How valid are KPIs in spare parts SCM performance measurement for a case company?
A KPI should support managerial control, so that issues can be identified and resolved. Companies produce numerous KPIs but there is a need to determine their
validity within a spare parts SCM PM system and thereby, to assess its capacity to provide useful information for managerial control in a case company.
1.6 Research Methodology
The research methodology is developed to address the three research questions and incorporates ideas generated from the literature review. An analytics approach is used to test the validity of KPIs in a case company’s PM program for spare parts SCM.
The first question, “Can the use of an industry standard framework enhance the performance management process of spare parts SCM for a case company?”, is dependent upon a theoretical overview of performance measurements used in SCM.
These will be reviewed to identify key concepts in PMs, while the analysis of significant papers identifies criteria that need to be considered in an effective KPI.
These factors will feed into the quantitative analysis section.
To answer the second research question, “Can the use of business analytics, including the use of “Big Data”, and new technologies, such as geographic information systems, enhance spare parts SCM process for a case company?”, new software technologies available in the field of geographic information systems will be used. This software will utilize Big Data leading to spatial analysis of the variables, which will increase the understanding of the data.
The third research question, “How valid are KPIs for spare parts SCM performance measurement in a case company?”, a systematic statistical approach as outlined by Oberg et al. (2016) will be applied. The statistical validation entails first finding and selecting the KPIs that will be used as variables. The selection criteria used is derived from the literature review, based predominantly upon de Leeuw and Beekman’s (2008) findings. Following the selection, the newly constructed variables are validated with the key statistical techniques identified by the literature review.
1.7 Research Structure
The structure of the thesis follows a logical progression. The initial chapter provides the background to the research including the focus and key concepts, motivation, research aim, literature review, research questions, methodological approach and outline. The Methodology chapter discusses a single case approach, the selection of a SCPM framework, and a three-phased research approach and the required data. A brief discussion of both Big Data and geographic information systems is included as both are found to be supportive of the data analysis. The next chapter introduces the specific case outlining the single case approach, details on the case company, its Global Spare Parts unit and the company need. It goes on to provide data sources and data analysis based upon the SCPM framework. The Discussion chapter summarizes the results in relation to each of the three research questions. The final chapter, Conclusion, discusses the benefits of using a business analytical approach to enhance the validity and utility of PM for both managers and academics. It also overviews the contributions of this thesis in the research area of spare part SCM and to the case company as well as limitations and future areas of research.
2 METHODOLOGY
This chapter overviews the methodology discussing the single case approach, selection of a SCPM framework, and the three research phases being used with the case company.
2.1 Single Case Study
A single case is used to provide a practical insight into the use of PMs for a spare part’s company. According to Perry (1998) case studies permit an analysis of “how and why” something is done and Ying (2009) supports their use as a relevant approach for “investigating a contemporary phenomenon in depth and within its real- life context”. The method has been proven to be effective as demonstrated in recent studies (Kasi 2007; Lin & Ghodrati 2012; Beekman & de Leeuw 2008). Further Gibbert (2005) advocates the use of single case studies as a highly valuable approach in the area of applied research.
The case company and its spare parts unit are representative of groups around the world that are responsible for spare parts SCM as they recognize the need and potential utility of a valid measurement but do not consider that their current process is effective. Consequently, this study provides the opportunity to support a company as well as the broader needs in spare parts SCM. Thus, it is hoped that the approach of applying a SCM PM framework, within a single case, will provide practical insights into managerial control of spare parts SCM for this case company as well as provide insights into the broader context of spare part SCM.
2.2 Supply Chain Performance Measurement Framework
The SCMP framework produces metrics that are indicators of the performance within a supply chain. There are numerous frameworks and therefore selection of the appropriate SCPM framework to be used in this research is critical. This requires an
understanding of its merits as well as those factors that are critical to represent a specific supply chain. Several frameworks were identified in the literature review. As this is a critical component of the research an expanded literature search was done which built upon the results of the literature review (Appendix 2).
The Value Chain Reference Model (VCOR) and Value Reference Model (VRM) (Domenico 2008) both successfully identify the critical success factors of each key business unit that need to be implemented in a variety of industries (Kirikova 2013).
However, the frameworks lack practicality in implementation due to the lack of operational metrics.
Other approaches that succeed in implementation are considered to be both the Multi-Criteria Decision Making Model (MCDM) and the Fuzzy Set Theory (Gurrea 2014). Both methods allow for strong metric development with concrete guidelines.
However, their lack of ease of use and understanding of the mechanics of implementation make their results open up to interpretation. An easier system for SCPM is considered to be the Economic Value Added (EVA), which focuses on the net profit subtracted from the equity cost of the company’s capital (Hofmann, 2010).
However, the narrowness of EVA’s scope to strictly focus on financial aspects limits its utility to specific environments.
The Balanced Scorecard (Norton 2011) and Supply Chain Operational Reference (SCOR) models (Georgise 2012) are methods that have all the strengths of the VCOR, VRM and EVA, plus ease of use and clearer interpretation of results than either the Fuzzy Set model or the MCDM. The Balanced Scorecard utility is limited given the predominantly qualitative nature of the results, whereas, the SCOR method is acknowledged as one of the most commonly used PM frameworks in SCM (Balfaqih 2016), as it effectively assesses both monitoring and controlling. It enables users to address, improve, and communicate SCM practices within and between all interested parties in the extended enterprise (Poluha 2007). Based upon this review
the SCOR framework is selected for use in this thesis, placing the other frameworks outside the scope of this research.
2.3 Three Research Phases
The methodology generally follows the three phases outlined by Demirkan and Delen (2013): descriptive, predictive and prescriptive. Figure 2 overviews this methodological approach as it addresses the research questions and the flow of information between the three phases.
Figure 2. Methodological approach to address research questions
2.3.1 Phase 1: “As Is” Description
The first phase is mainly concerned with a description of the “As Is” situation. The objective is to statistically characterize the important elements across the spare parts SCM that impact the PMs in the case company and to benchmark the various performance metrics. This addresses research question 2,” Can the use of business analytics, including the use of Big Data and new technologies, such as geographic information systems, support the process of developing useful performance metrics for a case company?”
Given the complex nature of spare parts SCM, this descriptive analysis is a critical step to ensure that the needs of the company are clearly understood before proceeding to the next phase (Hazen 2016). Also, this initial measurement is important to clearly understand the current company situation and therefore, to be able to later measure if any changes would produce positive results. With this clear understanding of the current state, it will be possible to conclude whether changes would enhance the company’s situation.
This involves basic statistics, testing data validity and exploring spatial distribution patterns of the data. This is particularly important in spare parts SCM as the systems can be very complex in terms of the number of processes as well as the amount of data that are inherent in the process (Huiskonen 2001). In this context, Big Data has the potential to address the complex need for information requirements in spare parts SCM. (Vera Baquero 2016) Additionally, the characterisation of the data is relevant to ensure that the analytical techniques of the second phase can be undertaken. The assessment of spatial trends, via the use of new technology, Geographic Information System (GIS), adds a new dimension yet potentially important aspect to the analysis of performance measurement. (Foote & Lynch 2015) These two aspects of data in spare part SCM, contribution of both Big Data and spatial patterns provide a challenge and opportunity that will be investigated to ensure that the selected framework is based on appropriate data.
2.3.2 Phase 2: Metrics’ Relationships
The second phase examines the relationship between the performance measurements and the impact of various variables on it as well. The basis of this phase is from Velimirović’s (2011) work examining how empirical data can be used for calculating KPIs in spare parts SCM. The purpose of this phase is to determine the significance of the association between the various performance metrics and in particular what factors provide the best insight into company performance as
measured by their KPI. This is based on undertaking the appropriate correlation, regression and principal component analyses that follow from Phase 1. This analysis will support a thorough understanding of spare parts SCM dynamics. This addresses research question 3, “Can the use of an industry standard performance measurement framework enhance the control of spare parts SCM for a case company?”
A key component in this phase is the calculation of the KPI. In the SCOR framework there are over 150 metrics available for a company to use which measure different aspects of a company’s performance. The decision as to which metric to use is guided by the needs of the specific company, including: company type, strategy and data availability. The critical criteria required for spare parts SCM PM are presented in Table 2. (de Leeuw and Beekman 2007) These criteria are used to filter through the SCOR framework and to identify the specific metrics that best suit for the case company needs and their capabilities.
2.3.3 Phase 3: Performance Enhancement
The third phase is the evaluation of the performance metrics to determine their validity. This phase is based on Oberg et al. (2016) requirement for quantitative methods and interpretations to provide increased credibility. This phase reviews the strengths and weaknesses found in the company’s current metrics and the new SCOR metrics and provides recommendations for incorporating enhancements into managerial practice. It is the critical stage in the operational aspect as it requires user understanding and acceptance of the metric. This phase addresses research question 1, “Can the use of an industry standard performance measurement framework enhance the control of spare parts SCM for a case company?”
Two elements of this phase are benchmarking and metric validation. Benchmarking of the current metrics is provided to support future investigations of performance measurement in the case company. Validation incorporates a statistical evaluation
and a graphic representation to facilitate managerial understanding as suggested by Oberg et al. (2016). Overall this will support evaluation of the utility of a KPI for managerial control.
3 CASE: SPARE PARTS SUPPLY CHAIN MANAGEMENT OF A MULTINATIONAL COMPANY
This chapter starts with a presentation of the case company, its spare parts unit and the company PM needs. Then it discusses the methods for data collection, data analysis, geographic information analysis and the calculation of the KPI using SCOR methods.
3.1 Case Company: KONE Corporation
The Finnish based company KONE Corporation, was founded in 1910 and is headquartered in Helsinki, Finland. It is an international engineering and service company employing approximately 49,000 personnel worldwide and in 2015 had annual revenue of 6.9 billion euros (KONE 2015, 14). The sales of the company are expected to grow by 2 – 6 percent for 2016 (KONE 2015, 11), showing the company to have a healthy level of growth. The firm is one of the largest manufacturers of elevators and escalators worldwide, and also provides maintenance services. In KONE’s 2015 Financial Report, a directive for improvement was for the development of better measures to mitigate the risks related to the SCM by analyzing and improving the processes (KONE 2015, 9). KONE’s focus on the development of better SCM is aligned with the aim of the research.
3.1.1 Global Spares Supply unit of the Case Company
The Global Spares Supply unit is the focus for the research. The Global Spares Supply is responsible for the management of tens of thousands of spare parts for a constantly increasing number of elevators, escalators and building door materials.
The unit’s mission statement is, “Right part, at the right time, to the right place, cost effectively, and at optimized inventory cost”. Striving for this mission is achieved through controlling all aspects of their vast material base. This creates a clear need for sophisticated business processes, data analysis and action tools for SCM. An
overview of the basic the Global Spares Supply unit information is included in the Figure 3 below. The figure expresses the size and international nature of the division.
It has production factories located in different continents with employees and customers working all over the globe. This clearly necessitates a PM system to be capable of capturing the sophisticated business environment that KONE Global Spare Parts operates. (Paakki 2007)
Figure 3. Overview of the GSS offices and performance levels for 2015
PMs were developed to monitor the Global Spares Supply’s internal processes to control their SCM. The control of these spare parts has been transformed over the past decade from reactive in nature, to proactive (Paakki 2007). Recently, in Global Spares Supply, the operative side of the inventory management process has been an area of focus. The scope of interest concerns the entire operational SCM and the inability of managers to understand and use their most important KPI or Core Performance Indicator, CPI. For KONE, there is a hierarchy of performance measurements and the CPI refers to a high-level performance indicator to which the lower level indicators should report. The CPI is Extended Service Performance (ESP).
At the center of this measurement is customer satisfaction, more specifically, whether
KONE performed in accordance to the contract agreement and external factors did not provide any impediments. It is the fundamental measure used by senior managers to assess performance and as a critical measure, it should relate to the lower level performance metrics.
3.1.2 Defining the Business Structure in Global Spares Supply
The SCM for Global Spares Supply is illustrated in Figure 4. The unit is composed of four teams: Inventory, Logistics and Invoicing, Material Management and Customer Service. The teams work together to ensure that the customer gets the “right part at the right time” for a reasonable cost. (Paakki 2007)
Figure 4. Supply chain flow for the four units within Global Spares Supply
The major functions of each of the groups are presented in Table 5 (Section 3.2.1).
The process of supplying parts is initiated via the Logistics team who handles delivery of the requested action. They liaise with the external vendors who produce the parts.
The invoicing function is found within this group. Customer Service works at the front
end by communicating directly with the customers to resolve issues and provides the interface with the Material Management team who facilitates the flow of materials to the appropriate vendor.
3.1.3 Case Company Needs
For the case company the issues of SCM for their spare parts is a challenge given the volatile and changing nature of spare parts SCM with the added complexity of the company’s global profile, leading to challenges in controlling actions. Within the Global Spare Supply unit, managers recognize the need and potential utility of valid PM process but are concerned that their current process is not as effective as it could be. Specifically, there is a concern that the current PMs are not meeting their needs for timely detailed information. There is a desire within the company for more timely information that would facilitate managerial action to be taken. Thus, a model or method for spurring the creation of a tool that can aid in managerial control of spare parts SCM, would be a desired development for managerial control. Consequently, this study provides the opportunity to support the case company.
3.2 Data Analysis
This section provides the details on how the data from the case is first compiled and then analyzed. The goal of this data analysis is to get reliable and pertinent information regarding the case to support the calculation of a KPI. The final part provides the details on how the KPIs are calculated using the SCOR framework as well as how successful the KPIs would be for managerial control.
3.2.1 Data Compilation
The first step in the analyses is the compilation of data from the three data sources, Servigistics, SAP and Qlikview. These data sources are required to provide the full spectrum of data required in the quantitative analysis of this research. The primary data source is the ERP system, which is the database engine that added the “Big
Data” aspect to the study. Servigisitcs supplemented the ERP system with pre- analyzed data providing more than the transactional information from SAP. Finally, QlikView is used as the hybrid server that utilized information from both servers and acted as a raw data validation tool. The data are extracted from the three sources:
Servigistics (assures data quality), QlikView (validates findings, provides master tables explaining columns and fields) and SAP (data warehouse, most observations are pulled from this source).
All of the data are loaded into a local data warehouse (SQL server). The server contains 1,190,377 unique observations with 57 column traits, meaning that a column based Database Management System is required. The columns contain several different data types, making data cleaning a necessity after pulling the data from the three data sources. A cleaning process for 2016 data is developed to check for missing data, duplicated data, and erroneous coding. In this process the data are updated in near real time.
A relational database is created on a local SQL server and is populated with the Big Data referenced previously. A relational database is structured to identify and organize relationships between stored lines of data. The relationships are formed to provide the SCPM quantitative tools, as outlined in the methodological approach. The resulting database is shown in Appendix 3 and colour maps each table to their originating source. Each table contains its title, which describes the tables’ contents, and the type of information contained in the table, example “raw data” means all field in table contain “raw data”. These tables also identify if the relationship between the tables and fields are: one to many, many to one or many to many relationships. Many is represented with the infinite symbol (∞) and one represented with the number 1.
(Peter 2011)
There are three different types of tables used in the data model. The model has transactional data tables which contain information that frequently changes. Next is the master data which contains more static data. Finally, there is a base table where
all queries occur. The queries’ outputs form the data to be inputted into the GIS and SCOR framework. These models are where the data science portion of the analytics takes place. Overall, this relational data model allows for the SCOR metric data and their logic, as listed in Figure 6 (see section 3.2.5), to be created and to be analyzed.
(Peter 2011)
Based on this data compilation process, the data compiled for the SCOR calculation meets the criteria of the three V’s requirement for Big Data. The first V “Volume” of big data is present, as there is an excess of a million observations of data that comprise gigabytes of information. “Velocity” is shown with the near real time presence of the data. “Variety” is displayed in the data with each observation displaying over 57 traits of different data types.
3.2.2 Statistical Characterization
The basic statistics of the PMs include mean, standard deviation and quartile measures. Several tests assess data distribution and data validity, particularly data suitability for more complicated statistical analysis. This includes: histogram plots, Anderson-Darling test for normality, as well as Bartlett tests for homoscedasticity of variances.
After completion of the basic statistics, more rigorous analysis of the data is done to identify the relationships between the performance metrics, in particular the relationship between the KPIs and the CPI. In addition, data are analyzed to identify the sources of variation to determine the characteristics required in the performance measurements. The testing of correlations is done using Pearson Product Moment Correlation and a Principal Component Analysis (PCA). The Pearson Product Moment Correlation indicates the presence of the relationship between the performance measurements. The PCA is a multi-variate non-parametric technique, which provides insights into the relationships between the metrics. It also provides a
roadmap for how to reduce multifaceted data set to lower dimensionality. This can reveal underlying factors, which are responsible for variation in the data. It accomplishes this by converting a set of observations into a new dimension of linearly uncorrelated components.
The first principle component is produced from the linear combinations of the x- variables that explain the most amount of the variation. The second principle component accounts for the most amount of residual variation in the data with the constraint that it is independent of the first component. This continues with the result being a number of components that are independent of each other and the influence of each component on each of the variables is presented. Typically, a significant amount of the total variation is explained in the first two to three components.
However, this is not always the case. A challenge with the resulting components is to identify what they mean, since it is not always clear from the results. To support the interpretation of the factors a series of tools are used; three dimensional plots of the principal components and an understanding of the metrics that are the basis for the analyses. Finally, linear regression and multiple linear regression analysis are conducted to determine the predictability of the CPI metric based on the other performance metrics. The statistical analysis uses Minitab version 17. (Shlens 2017)
3.2.3 KPI Calculation, Global Spare Supply Unit
In the Global Spare Supply unit, 12 commonly used KPIs are listed in Table 3. This includes 3 KPI’s in each of the four units with each representing a critical stage in the SCM for that team.
Table 3. Twelve key performance indicators used in the Global Spare Supply unit GSS UNIT Inventory Customer
Service
Invoicing Material Management
Key
Performance Indicators
Excess Stock (ISCO 1)
Plant Emails (CS1)
DC Performance (IN1)
Missing Order Conf (MM1) Forecast Accuracy
(ISCO 2)
Department Emails (CS2)
Freight Debit Memo (IN2)
Late Back Orders (MM2) Stock Rotation
(ISCO 3)
Distributor Emails (CS3)
Freight Cost Accuracy (IN3)
Future Back Orders (MM3) Each of the KPIs are calculated in a unique manner using internal data considered pertinent to that unit. The sampling dates are not consistent between the units as 50 percent of the metrics are measured on a weekly basis while the other 50 percent are calculated on a monthly basis.
3.2.4 Geographic Information System Analysis
The main role of the Geographic Information System (GIS) is to display the results from the analysis in a spatial/map format. In this case the performance results as well as shipping conditions for each country are initially calculated in the SQL server and then exported as comma separated values (CSV). Finally they are appended onto the attribute table within the GIS.
The GIS software used is QGIS (version 2.16) which is an open source freeware package that started in 2002 (QGIS 2016). The GIS data for the project is taken from the Natural Earth web site, which provides free downloads of digital data (Natural Earth Science 2016). For the purposes of this thesis, two data layers are included, both using the WGS84 projection. The first is a set of 176 unique polygons representing the various countries and independent states around the world. The data includes over 45 different attributes, such as: population, level of economic development, etc. For this study the country name is a key label as it permits the data from the case company to be appended and thereby rendering it available for geographic analysis.
The other data set is a raster image of the entire globe referred to as “Natural Earth 1”
(Natural Earth Science 2016). The image is mainly used for display purposes.
However, it is also used as a base map to enter polygons for countries or independent states that are not included in the Natural Earth country data base. The eight countries/independent states that are added into the country file were: Aland Islands, Andorra, Faroe Islands, Country K, Country C, Jersey, Malta and Mauritius.
3.2.5 SCOR Metrics Calculation
The SCOR framework is chosen to test the capacity to develop a metric that would be a framework for building future performance metrics. The process for developing the metric follows a rigorous guideline, provided by SCOR, that occurs in three steps (Figure 5). First, one of the many metrics within SCOR needs to be selected. Once the metric is determined the appropriate data to populate the model are identified.
Then the final step is extracting the relevant data from the appropriate data base. A series of scripts and queries “pull” data from the databases identified in section 3.3.2.
This is complicated by the Big Data nature, which requires a server capable of handling large data volumes.
Figure 5. Three critical process steps in calculating SCOR metrics and preparation for Data Science
The study uses Transact-SQL (T-SQL), which is a branch of a SQL Server. An SQL Server is a business analytics tool capable of storing, processing and extracting big data. All applications that communicate with an SQL Server are able to be sent to the server via utilizing T-SQL statements. These statements translate code into actions that are carried out by the SQL Server. The T-SQL server collates the data and prepares the data for subsequent data science analysis (Microsoft 2017). The data science allows business analytics to commence and test the research question and its hypotheses. The sequence of steps, in calculating the SCOR performance metric and transfer to the data science analysis, is presented in section five of the results.
The selection of the appropriate SCOR metric is difficult as there are over 150 possible metrics and these depend on the specific industry requiring the metric. There are three categories of metrics from which the metric can be selected: Reliability, Responsiveness and Agility (Figure 6). To support a holistic approach at least one of the five metrics occurs in each of the three categories. The goal was that these five metrics would provide an adequate sample size to relate the SCOR metrics to the case company’s CPI (ESP). The metrics are also selected to align with the criteria identified by de Leeuw and Beekman (2007) and listed in Table 2. Three criteria are covered by the SCOR metrics selected: Stock turnover rate, inventory process time, total stock turnover, cycle time (1); Average service level, stock accessibility, fill rate time (4) and Punctual delivery performance at requested time and place (6).
Figure 6. Characteristics of SCOR metrics as calculated in study
The five-metrics chosen are: Fill Rate, Perfect Order Fulfillment (POF), Order Fulfillment Lead Time (OFDT), SC Dwell Time and SC Process Time. An outline for the calculations is given in the calculations column, followed by the next column providing a broad definition of the goal for the SCOR Performance Metric (Table 4).
The generic calculations definition is sufficient for all of the metrics except for Fill Rate.
The Fill Rate requires additional information regarding the normal delivery time (days) for a stock item and a category specification for the item. According to the case company’s policy, 24 hours is tailored to define a normal delivery time from stock and the item category definition for each item is used to classify stock material.
Table 4. SCOR Metric Composition
Attribute Metric Calculations SCOR Metric Definition
Reference from SCOR Metric Book
SC Agility
Fill Rate Mean [End Customer GR]
- [GI Date]) < 24 hours)
% of stock orders delivered within 1 day of order.
- DI 2. Customer Delivery
SC Response
Perfect Order Fulfillment
Average per Order ([Total Perfect Orders] / [Total Number of Orders])
% of orders supplied on time & in full with no flaws.
- RL.2.1 % Orders Delivered in Full - RL.2.4 Perfect Condition
- RL.3.19 % Orders Received Defect Free
SC Reliability
Order Fulfillment Lead Time
Average per Order ([Sum Actual Cycle Times For All Orders Delivered] / [Total Number Of Orders Delivered])
Days from order receipt to customer receipt.
- RS.2.2 Make Cycle
Time - RS.2.3 Deliver Cycle
Time - RS.3.96 Pick Product
Cycle Time
SC Reliability
SC Dwell Time
Average per Order ([Date Goods Issued Created] - [Date Order Created])
Time a customer allotted to business before action is required on order
- RS 1.1 Perfect Order Fulfillment
SC Reliability
SC Process Time
Average per Order ([End Customer GR] - [GI Date])
Time by operations to carry out the order
- DI 2. Customer Delivery
The calculations from SCOR provide the metrics as well as a number of key characteristics of the processes including: “Metric” (Name of metric), “SCOR Performance” (the performance of the SCOR generated metrics), “Metric Correlation”
(fit of the case companies KPI correlation to the CPI) and finally “Metrics Normality”
(how stable the metrics overall process performed). These findings are presented in section 4.4 of the results.
3.3 Validity and Reliability of the Case
To measure the success of an empirical measurement there are two commonly used properties: reliability and validity (Carmines & Zeller 1979). Although there has been refinement in the use of the terms since their monograph, the concepts remain basic to any study involving measurement. Reliability is a measure of the repeatability of the results and validity refers to the degree to which you are actually measuring the phenomenon that you set out to measure (Mason 1999; Ying 2009). A successful measurement must possess both qualities. However, it should be appreciated that something can be reliable without being valid; whereas valid measures must be reliable. The terms can refer to any empirical measurement and in this section. They will be used to address the measurement of KPIs that are developed using the SCOR method for the case company. (Carmines & Zeller 1979)
3.3.1 Reliability
Reliability, as introduced above, is a measure of repeatability; if the same results can be found when the work is repeated under similar conditions. The test for reliability needs to identify those errors in measurement that are related to the measurement process versus those which are due to variability in the true score. The variability in the true score would be replicated in other situations and a reliable measure would mainly be impacted by this variation. To provide reliable results the data are assessed for normality as an indicator of the stability of the population. Also, documentation of the data analysis, including KPI calculation and data preparation, needs to be clear. (Ying 2009)
3.3.2 Validity
Validity, as introduced above, measures accuracy and is the basis for confidence in results. In this case the validity would be a measure of the degree to which the indicator mimicked the results of the performance of the case company as measured by the CPI. The situation of a KPI having the status “green” (positive) while the CPI having the status “red” (negative) needs to be avoided. There are several steps taken to ensure the validity of the KPI. First an appropriate framework for calculating a KPI based on accepted theory and alignment with company strategy ensures that the right process is being assessed. Then testing of the data including tests of normality ensures a robust database. Finally, correlation testing between PM is required to provide the degree to which the new measurements relate to the performance of the company. This is referred to as criterion validity. (Ying 2009)
The validity of extrapolating the method to other components of the case company or beyond is not tested. This has to be done to further validate the process and it is expected that if the same rigor is used in data preparation, similar results will be found.
